1. Introduction
Cardiovascular disease (CVD) is a major contributor to the growing public health epidemic in non-communicable diseases (NCDs) [
1]. CVDs account for the most NCD deaths (56.5 million), equating to 18.5 million people annually, followed by cancers (10.0 million), respiratory diseases (4.0 million), and diabetes (2.9 million) [
1]. The majority of deaths occur prematurely at an age below 70 years [
2]. Alarmingly, most of these deaths are in low- and middle-income countries (LMIC) [
3]. CVDs could be referred to as unique conditions of the heart and blood vessels that are essentially and generally caused by common exposure to a trait of a person’s character, defined as a risk factor [
1,
3]. Among these sets of risk factors are those attributed to habitual behaviors, such as smoking, and metabolic factors, such as hypertension. Furthermore, these risk factors can also be linked to each other, particularly smoking as a risk factor for hypertension [
3].
Hypertension is considered one of the dominant risk factors of CVDs, with approximately 1.13 billion individuals exposed to this disease globally. Studies show that hypertension remains undiagnosed in many individuals, especially in sub-Saharan Africa, due to the disease’s hidden symptoms [
4,
5,
6,
7,
8,
9]. In later life, this condition can directly or indirectly lead to premature deaths owing to the complexities that develop in untreated patients or as a result of late diagnosis [
4,
10,
11]. The most common health impacts resulting from high blood pressure include acute myocardial infarction, strokes, cardiac failures, and renal failure, among others [
12,
13,
14,
15].
Research findings indicate an escalating rate of high blood pressure mortalities in LMICs, a rate that is equivalent to approximately twice that of high-income countries. For instance, 7 and 25% of individuals in high-income countries and Africa, respectively, are likely to die before their 60th birthday due to high blood pressure [
11,
16,
17,
18,
19,
20]. A systematic review on sub-Saharan Africa also indicated similar findings with varying levels of evidence on the prevalence of high blood pressure in the region, generating a median of 29% [
19]. Nevertheless, continuous evidence obtained from various regions indicates a high prevalence of hypertension (between 9 and 50%) among urban settlers and in association with rural–urban migration. Moreover, these findings have been reflected in certain Kenyan studies [
21,
22,
23].
Smoking has been described as a major global risk factor for various CVD conditions, including stroke, peripheral vascular disease, coronary heart disease, hypertension, and cancers, among others [
24,
25]. Studies have reported that from 63 to 68% of NCDs attributed to death are highly associated with tobacco [
26,
27]. Nevertheless, a number of deaths occur below the age of 60, with a higher percentage in developing countries than the percentage of deaths in higher-income countries of approximately 29 and 13%, respectively. Furthermore, an estimated 12.7% of the 7 million tobacco deaths are due to secondary smoke exposure [
28].
Smoking and hypertension individually pose alarming consequences, and this concern is heightened when they are both associated with the same individual [
29,
30]. This particular modifiable risk needs to be better understood, allowing the development of suitable control measures, which may result in improved health and reduced mortality and morbidity due to the elimination of the double shared risk it poses. Information on ill health within sub-Saharan Africa is scarce but is not different from the situation found in Kenya [
31]. To our knowledge, no study has examined the relationship between smoking and hypertension indices in Kenya using a large, nationally representative, and population-based data set. Accordingly, this study examines the association between tobacco use and hypertension in the Kenyan population. The study further controls for BMI, waist–hip ratio, alcohol consumption, physical activity, and socio-demographic or economic factors to establish the prevalent link to hypertension among the Kenyan population. The research questions posed are: how is tobacco use associated with hypertension, and what are the risk factors in hypertensive individuals in Kenya?
3. Results
This study aimed to investigate the association between tobacco use and the development of hypertension as well as other known hypertensive risk factors in Kenya. The population was characterized as follows:
3.1. Characteristics of the Study Population
The study participants included in this study ranged between 18 and 69 years old, with 58.8% of all participants identified as females (2559) after excluding all pregnant women. Moreover, the number of women in all the rural–urban settings was high within each of the wealth quantiles, as shown in
Table 1. Each quantile represented approximately 20% of the participants. The participants’ mean age was 37.82 ± 13.46 with almost 70% below 44 years, and about 67% were married, and 40 and 9.9% were self-employed or unemployed, respectively. Homemakers made up 24.5% of participants, and at least 47% had completed their primary education or 7.7 ± 5 years of school. In general, 51% of all participants lived in a rural setting across the country. Among the hypertensives, 36.6% were males, and 51.9% were urban dwellers. The difference within the groups was not significant at the 0.05 level. Within the marital status and age groups, approximately two thirds of the hypertensives were married and between 30 and 59 years of age, while the self-employed and homemakers were also more hypertensive within the occupation status at 40 and 24%, respectively. In the wealth quantile groups, the poorest had the lower percentage (13.5%) among hypertensive, while the middle, fourth, and fifth (richest) quantiles were higher at approximately 22% each.
Higher proportions of the demographic categories among current tobacco users were identified in males (79%), in the age groups between 30–44 (41%) and 45–59 (27%), the married (64%), the self-employed (43%) and employed (21%), and the rural dwellers (56%). In the wealth quantile category, an estimated one fifth were associated with the second and middle quantiles, whereas most of the current tobacco users were the poorest (30.6%). Among the former tobacco users, similar groups as those in the current tobacco users had higher numbers presented as quitters except the wealth quantile groups where the poorest were indicated as having the lowest percentage of quitters at 12.8%.
Table 2 lists the descriptive statistics of physical measures and continuous variables. The results indicate that the two parameters, systolic and diastolic blood pressure, had exceptionally low and extremely high markers (diastolic 48–152 mm Hg and systolic 71–264 mm Hg) with a mean of 83 ± 12 and 127 ± 20 mm Hg for the diastolic and systolic blood pressure, respectively. These two parameters also varied among males and females and were found to have a greater degree of skewing to the right among two age groups of 30–59, as illustrated in
Figure 1. The mean of the waist–hip ratio (WHR) of participants is above the recommended normal ratio for women at 0.8±0.08 while the BMI mean is just below the overweight threshold at 23.5 ± 5.08 kg/m
2. The fasting blood glucose mean was reported at 4.7 ± 1.4, and the mean of physical activity was noted to be 6 ± 4.2 h per day with at least no hours of physical exercise for others.
3.2. Prevalence of Risk Factors among Participants
The prevalence of hypertensive participants was reported as 22%, as noted in
Table 3. Moreover, the findings indicated that tobacco consumption and former tobacco users among the participants were 12.6 and 8%, respectively. More people were exposed to alcohol drinking than tobacco use, with a prevalence of 21%. Furthermore, obesity levels were reported as high, based on BMI (obese 10%, overweight 21%) and WHR (obese 25%, overweight 26%). However, the raised blood glucose levels and physical activity levels of less than 30 min were both found to be relatively low compared to all other risk factors, and both were reported as 6%. The highest prevalence was found in fruit and vegetable consumption of less than five servings, which was reported as 95 and 96%, respectively (see
Table 3).
3.3. Prevalence of Risk Factors among Hypertensive Participants
Among the hypertensive females, 4.3% were current tobacco users, compared to 23.4% of males. Approximately 50% of men were found to consume alcohol compared to only 8.7% of females. Additionally, 41.3% of males and 25.2% of females were exposed to second-hand smoke. Most males had a WHR of <0.9 (59.4%), or they were overweight (34.3%), while most females were obese (53.4%) and overweight (26.3%) when classified according to their WHR. However, when the BMI classification was used, only 25.8% of females were obese, and 33.2% were considered overweight, while 22 and 7% of males were classed as overweight and obese, respectively. Most of the males were found to have a normal BMI (48%). All the differences were found to be statistically significant in all the categories mentioned above except for a few factors. First, the raised blood glucose was found to be high (13.1%) among females compared to males (10.6%). Both females and males were reported to have low vegetable consumption, low fruit consumption, and less than 30 min of physical activity per day.
Table 4 lists the health practices noted in individuals that are also indicated as hypertension risks. These health practices could lead to self-awareness and an understanding of the distribution in healthcare facility utilization. The major primary source of healthcare for most Kenyans is a dispensary or community health worker (34%), followed by referrals to public hospitals (28%), which were previously referred to as district and provincial hospitals. This seems to be the preferred source in almost all categories of the explored risk factors. Specifically, over one third utilized dispensaries in each group, while more than a quarter utilized referrals to public hospitals. There is also a high rate of self-medication or use of alternative therapy (9%), which is also highly preferred among former tobacco users (17.4%), current tobacco users (14.2%), current alcohol users (13.8%), obese individuals (12.2%), and those considered WHR overweight (11%).
Of the participants, almost 51% had never been checked for blood pressure, and within this group, the majority of current tobacco and alcohol users had never been checked (71 and 62%), respectively. In contrast, over two thirds of those classed as obese and overweight, and those with raised blood glucose were more likely to have been checked for hypertension. Among those checked for hypertension, 50% of individuals with raised blood glucose were diagnosed to be hypertensive, while almost 25% consisted of individuals from the other risk factors (BMI, WHR, alcohol consumption, and tobacco use). Over 50% of the diagnoses had occurred during the 12 months prior to the survey. Nevertheless, medication adherence seems to be low, with approximately 60–80.8% not taking the prescribed medication within all the groups, and the highest non-conforming group being the past alcohol users.
3.4. Linear Regression of Systolic Blood Pressure and Diastolic Blood Pressure on Tobacco Use
The backward stepwise regression was used to establish a fitting model, and the one presented here was found to have a higher explanatory power (R2) and a low BIC score (40). Consequently, a comparative model was run with all the variables to determine the difference from the main model. The models were run on an unweighted scale first, followed by a weighted model. The unweighted bivariate model indicated non-significant results on current tobacco users and estimated 6.9 mmHg higher SBP in former tobacco users. These measures and statistical significances were altered after controlling for other explanatory variables and are now presented in weighted models below.
Using the weighted model and after controlling for the relevant variables, the association with the development of systolic pressure for current tobacco users was found to be 3.14 mmHg lower than the systolic pressure of never tobacco users. A 1.5 mmHg higher systolic mean had been predicted in the bivariate model, although it was deemed non-significant. On the other hand, former tobacco users’ low mmHg systolic pressure (0.76) means compared to never-tobacco users could not be justified statistically. This was despite the level in the bivariate model being observed to be 7.4 mmHg higher than that of never-tobacco users. In addition to current tobacco users, several other factors were associated with a higher SBP. These included ages 60–59 (22), 45–59 (11.6), 30–44 (1.9), males (5.9), never married (1.7) students (3.0), current alcohol users (4.1), BMI (overweight 5.0 obese 8.8) WHR (overweight and obese approximately 2.4), former alcohol users (3.3), or reduction of SBP: underweight (6.4).
A weighted bivariate and multiple linear regression model was then used to define the association between DBP and tobacco use. However, this model could not statistically confirm the low mmHg measure of DP in current users (0.76) and former users (1.15). Nevertheless, an association was identified between higher mmHg measures and DBP in the categories of age range 60–59 (6.2), 45–59 (6.4), 30–44 (2), current alcohol use (2.7), BMI (overweight 3.2 and obese 6.3), and WHR (overweight 1.7 and obese 1.3).
3.5. Multiple Logistic Regression Model
The association between tobacco use and the development of hypertension, as previously mentioned, was determined by logistic regression at a 95% confidence interval (CI). In a weighted bivariate logistic regression model, estimates indicated 1.48 (CI: 1.16–2.3) higher odds of hypertension in former tobacco users, while current tobacco users, even though non-significant, had 0.94 (CI: 0.77–1.44) odds. However, in a multivariate model, tobacco use remained non-significant with less likelihood of 0.12 (0.88–CI: 0.61–1.28) and 0.1 (0.99–0.64–1.53) current and former users, respectively, as observed in the bivariate model. Therefore, significant odds for the risk associated with developing hypertension were found and included age with those who were older found to have a higher risk. Specifically, the highest risk was reported in those aged 60–69 with odds of 5.3 (CI: 3.70–7.46), followed by the 45–59 age group with odds of 3.53 (CI: 2.6–4.80), and the 30–44 age group was found to have odds of 1.33 (CI: 1.00–1.76). Other identified risks included BMI >25 kg/m
2 with odds of 2.6 (CI: 1.9–3.6), WHR overweight with odds of 1.4 (CI: 1.07–1.7), obese with odds of 1.6 (1.25–2.09), current alcohol users with odds of 1.68 (CI: 1.24–2.3), and underweight BMI, which was associated with lower odds of developing hypertension (0.55) (see
Table 5). Thus, in this model, the constant of 0.072 odds (0.04–0.12) was linked to the base factors that included females of 18–29 years of age of normal BMI who never used tobacco or alcohol and were married. Moreover, it was also linked to the base factors that despite being poor, they had finished secondary school, were also employed, and had lived in a rural area with high levels of physical activity.
3.6. Sensitivity Analysis
The comparison models on the effects of the explanatory variables had minor differences from the main models. These were noted in the finding of associations between SBP and physical exercise that was less than 30 min a day of 3.8 mmHg lower than the systolic means of those who would undertake a physical activity for more than 60 min per day. Moreover, it was also identified in individuals who ate the recommended quantity of vegetables and had a lower (4.5 mmHg) SBP than those who did not consume five vegetables a day. Furthermore, a unit increase of cholesterol resulted in a 1.2 mmHg increase of systolic pressure, and for tobacco use, this resulted in a 3.7 mmHg lower level in current users, while the student effect was no longer significant as the net of all other variables (see
Table 6).
Conversely, DP was associated with a 4.6 mmHg lower diastolic means in those who were unable to work or had retired. Likewise, a unit increase of cholesterol was linked to a 0.9 mmHg increase in diastolic blood pressure, and a raised blood glucose level increased the diastolic means by 2.52 mmHg. Additionally, the indication of current tobacco users changed to signify a higher diastolic mean of 0.25 mmHg which was considered not significant, while the WHR obese effect of 1.3 mmHg was reduced to non-significant.
In the logistic model, a raised blood glucose was linked to higher risks of hypertension with 1.6 odds, while the effect of the 1.3 odds on 30–44 years old was not significant. The coefficient and odds ratio changes ranged between ±0.02 to ±1.5 in all variables. Nevertheless, the comparison models analyzed the effects using a smaller sample (n = 3057) than that employed in the main model sample (n = 4272). This was due to the exclusion of participants with missing data for some variables, as the analysis used complete case analysis.
4. Discussion
In our first research question tobacco use could not statistically justify the protective factor among the current users for all the study outcomes (systolic, diastolic and hypertensin). While former tobacco users were likely to be hypertensive only in a bivariate perspective, which does change by controlling for all the other factors though not statistically significant. Among the other factors in our second research question BMI, current alcohol use, waist–hip ratio, and age were significantly associated with hypertension. We discuss further these estimates and other findings below.
The overall prevalence of hypertension in men and non-expectant women within the age range of 18–69 was 21.7%, which was below the global level and that of LMIC (31.1%) in accordance with the 2010 estimated prevalence (8). The prevalence was also well within the range of the various studies conducted in different settings in Kenya (6.7–50%) and sub-Saharan Africa (6–48%) [
4,
42,
43]. The differences in the prevalence in the various studies could be attributed to the population’s heterogeneity and the variance in behavioral risks within these groups. For instance, a study in Kenya that reported a prevalence of 6.7% was conducted in an urban setting among males over 15 years who were returning from a mosque. In addition to their cut-off point (>140/>90 mmHg), this study undertook a subjective decision in classifying individuals as either hypertensive or not, and this subjectivity could have affected the results [
43]. On the other hand, the highest prevalence was attained in a cross-sectional study that was conducted on individuals who were 50 years or older and in an urban setting. Accordingly, such a setting would potentially result in a higher prevalence [
5].
Contrary to several studies that show higher prevalence of hypertension in men [
41,
44,
45,
46,
47,
48,
49], this study found higher hypertension prevalence in women (23.4%) than men (19.3%), which were similar to South African and Tanzanian studies that found females to have higher prevalence though South Africa prevalence was seen to decrease in women with higher education [
47,
50]. In addition, as in many other sub-Saharan countries, higher prevalence was among 60–69 years old 42%, the retired or those unable to work 38%, 45–59 year old 35%, the widowed 34%, and divorced or separated 27% [
20].
4.1. Hypertensive Awareness
Generally, hypertensive awareness in most African regions has been low. As indicated in the survey used in this study, more than 50% had never been tested for their blood pressure. Such a lack of testing makes it even more difficult to control the epidemic if such practices are not reversed. These findings were in contrast to studies conducted in regions of the Americas or Europe [
1,
3]. For instance, one study in Brazil had indicated full awareness in the entire adult population, with at least one fifth of participants who did not adhere to medication [
48]. However, in this study, the participants in the different risk factor groups were reported to have a lower knowledge trend of their hypertensive status. Among current tobacco users and alcohol users, an estimated two thirds were not aware of their status or had never been tested for hypertension. In addition, those diagnosed within this group were less likely to follow the prescribed treatment, with approximately seven in every ten failing to meet treatment requirements.
Furthermore, the findings indicated that most health workers were more likely to check hypertension in individuals who were either overweight or obese or tested for impaired glucose or had a raised blood glucose level compared to other risk factors. This could be due to the apparent knowledge of the blood pressure risk among these groups, even though the participant’s adherence to treatment remains below one third except the raised blood glucose group.
4.2. Hypertension Risk Factors
The prevalence of tobacco use was reported as 12.6% and does not differ greatly from the national survey findings [
32]. However, the results indicate that there was a higher prevalence of tobacco use in males (23%) compared to females (4%) among the hypertensive. Nevertheless, hypertensive females were also characterized as obese or overweight or with a BMI greater than 25 kg/m2 and were more likely to be exposed to second-hand smoke. Tobacco use was low in this group, but alcohol use was substantially higher (9%). Conversely, 50% of hypertensive males consumed alcohol, while 40% were exposed to second-hand smoke. These findings could potentially be explained by the reasoning that those who drink also smoke. Accordingly, those non-smokers who drink are more likely to be exposed to such second-hand smoke [
49]. In addition, they were also characterized as overweight, while a few were obese compared to women.
In the study, current tobacco use failed to meet researchers′ expectations of increasing SBP. Instead, current tobacco use was linked to 3.14 mmHg lower systolic means compared to never tobacco users. Furthermore, its effect on DBP could not be confirmed using the study data, and this was similar to the logistic regression results, as lower odds were observed in current tobacco users. The odds of current tobacco users did not differ when analyzed differently as smokers or smokeless users even after controlling for all other variables. Thus, as in previous studies that failed to establish a link of either smoking or smokeless tobacco with hypertension, this study also reported findings that were contrary to the findings in a study performed in Yala Kenya with odds of at least two among current smokers and former smokers. The Yala study was supported by other studies, such as a Rwandan study that found almost 1.5 odds in current smokers as well as former smokers, 2.3 odds in smokeless tobacco use of DP among rural Indians and Bangladeshis, and 3.5 odds for smokers, while tobacco use were only found to be significant in bivariate analysis [
8,
41,
46,
49,
50,
51,
52,
53]. Nevertheless, Green, Jucha, and Luz (1986) indicated that blood pressure may acutely be escalated due to smoking but shows a lower blood pressure among smokers than never smokers or former smokers [
54].
This study indicates the magnitude of the age effect on hypertension and diastolic or SBP. Hypertension was found to be even higher in the older population with odds of 5.3, a 6.2 mmHg higher diastolic means, and 22 mmHg higher systolic pressure in 60–69 year-old. In contrast, among the 45–59 age group, the magnitude was a 6.4 mmHg higher diastolic mean, an 11.6 mmHg systolic means, and hypertension odds of approximately 3.7 [
4,
44]. Additional risk factors that were associated with all the outcome parameters accessed included BMI, alcohol consumption, and WHR with at least 1.04–2.6 odds of hypertension, a 1.3–6.3 mmHg higher DP means, and a 2.5–8 mmHg higher means of systolic pressure.
However, participants who were male, past alcohol users, and single, and a student were also associated with a SBP with a 1.6–5.9 mmHg higher means while current tobacco use and an underweight BMI were identified as protective factors of SBP. On the other hand, contrary to several studies, living in urban or rural settings was not significantly associated with developing hypertension (systolic or diastolic) blood pressure nor was the level of wealth, education or the level of physical activity, and the marital status (divorced, separated or widowed) [
23,
41,
42,
55,
56,
57,
58]. Similar results were also found in Uganda with the exception of alcohol use [
59]. It is also worth mentioning that even though the results were not significant, they were consistent with the findings in the contradictory studies above. Hence, the outcomes in this study could have been affected by the size within some groups, which might have affected their statistical power to influence the results. For instance, within marital status, the married (analysis base) sample proportion was 67.2%, while the widowed (1.2 odds) or divorced/separated (1.1 odds) had a proportion of an estimated 7% each.
4.3. Strengths and Limitations
The data used in this study are the most recent and provide relevant information on demographics as well as behavioral characteristics of the sample participants from the national population. Furthermore, the sample in the study was proportionally balanced, especially with regard to gender, region, and wealth quantiles. Hence, the results of these variables had enough statistical power to affirm their specific relationship to the study outcomes. Thus, it is probable that the results that were achieved here are the true effect of the outcomes. Some limitations include the limited data on various dietary risk factors, such as salt, which is known to have a detrimental association with hypertension. Moreover, there was a larger proportion of missing data on cholesterol, fasting blood glucose, fruit, and vegetable intake, which led to the exclusion of results used in the main models of analysis. On the other hand, blood pressure measurements were not taken within recommend time hence conditions surrounding blood pressure variation might have affected the findings [
37].
4.4. Further Research
The findings in this study enrich the body of research into this area and can act as a cornerstone for further research that should consist of standard measures, especially regarding assessing the dietary salt aspect that was not analyzed in this study. Tests on interactions, though beyond the scope of this study, were not able to be predicated. However, the effect of tobacco use, alcohol, and physical exercise interactions could be studied to establish if either of them had an influence on pre-eminent the risk of the other.
5. Conclusions
This study reports the association between tobacco use and the burden of hypertension. The findings also illustrate an important burden of hypertension in the Kenyan population. Kenya is in line to combat NCDs and has developed various policies that endeavor to integrate existing national and global initiatives with the ultimate aim of reversing NCDs by 2020. Therefore, the findings of this study and those found in regional or specific settings in Kenya could be used as evidence in the prevention of hypertension and to support the equipping of healthcare facilities for screening those at risk. There is a need for greater awareness of hypertension in the general population and among healthcare givers. In addition, within the health system, greater emphasis needs to be placed on the detection, treatment, and control of high blood pressure. In conclusion, health policies in Kenya must subsequently account for the control of hypertension. Thus, health policies and health promotion programs should be developed at all levels for health promotion and prevention.